The present invention relates to an image processing apparatus used in an image forming apparatus such as a digital copying machine wherein an image of an original is read by image reading means such as a scanner, the read image is subjected to a predetermined image process such as an image quality process or an editing process, and the resultant image is printed out on a paper sheet by image forming means such as an electrophotographic printer.
In most conventional devices dealing with images, such as a color/monochromatic copying machine, a scanner or a printer, an image signal is expressed by 8 bits per pixel. The reason is that, in either of a case of treating an image by software and a case of treating an image by hardware, processing in an 8-bit unit is easy because of limitations of memory structure, and if more than 8 bits are used, double the number of bits is required.
For example, in the case of a color image, incident light is converted by a color image sensor to electric signals of three primary colors, R (red), G (green) and B (blue). In this case, 8 bits×3=24 bits are required. On the other hand, in the case of color-separated images, which are separated on the basis of a well-known subtractive color mixing process, that is, four-color images C, M, Y and K of yellow (y), magenta (m), cyan (c) and black (k), 8 bits×4=32 bits are required. In an L*a*b* color space, 8 bits×3=24 bits are required.
When an image is treated by a copying machine, there is a case where image information of several bits, in addition to 8-bit image signal processing, needs to be processed at the same time.
For example, in the case of the copying machine, in general, in order to obtain a high-quality output image, image areas of a character part and a photograph part of an original are separated and different image processes are carried out for the character part and photograph part (sharpness is given to the character part with important put on the resolution), and smoothness given to the photograph part with importance put on the tone). In this case, a process block for separating image areas is provided in the image process. The block for image area separation outputs a l-bit discrimination signal representing, for example, whether the object pixel is a character or a photograph. A filter block uses the l-bit discrimination signal and, where the l-bit discrimination signal represents “character”, the filter block performs a filtering process by selecting a filter parameter which is suitable for character reproduction and enhances the resolution. Where the discrimination signal represents “photograph”, the filter block performs a filtering process by selecting a filter parameter which is suitable for a photograph and smoothes the image. Thus, the suitable image process corresponding to the discrimination signal is enabled, and high-quality output images can be obtained.
However, if the above-described l-bit discrimination signal is used for each pixel, a 9-bit image signal is needed. That is, the l-bit discrimination signal is added to the 8-bit image signal. The use of the image signal of more than 8 bits makes the system design difficult and increases the cost because of the limitations of hardware and software. Even where several-bit additional information needs to be processed in addition to the 8-bit image signal, it is desirable that processing of 8 bits in total be carried out for reasons of hardware and software.
In order to solve this problem, there may be a method of compressing an (8+n) bit image signal to 8-bit signal. However, in an ordinary compression system, image information may be lost at the time of compression. In particular, loss of n-bit additional information is very serious in the system. Thus, it is not proper to compress an (8+n) bit image signal to 8-bit signal. Moreover, the compressing process itself increases hardware cost and contradicts the reduction in the number of bits of images. Therefore, there is a demand for the advent of a method wherein an (8+n) bit image signal can easily be converted to 8-bit signal and necessary information and an n-bit image signal can be extracted without loss.